Using Fault Tree Analysis (FTA) to analyze root cause

Picture of Adam Bahret
Adam Bahret
FTA with RCA

Whenever products fail causing the program delay or customer complaints or warranty returns, executives say: “Find the root cause of this failure and fix it”. While there are many methods to conduct root cause analysis, today I want to share how Fault Tree Analysis (FTA) can be used for this purpose.

Root Cause Analysis (RCA) methods are plenty – Fishbone diagram, 5 Whys, Apollo method. These are effective tools when the investigation team is starting new and all possible information must be considered. While this approach is ideal and recommended, in practice, there is a time-constraint and cost to acquire failure information is high. Also, when the system is complex with tens of components listing all the possible causes will create a fishbone diagram that is as big as the Great Blue Whale.

  1. Is there a way to incorporate existing engineering knowledge to hasten this process? 
  2. How do we factor in the interdependency of causes?

Answer to both the questions is FTA. Fault Tree Analysis (FTA) is a systematic, deductive method for defining the occurrence of a single undesirable event and finding all the possible triggers for that event. It uses logical gates to represent the interaction of causes and its higher effect.

Sample Fault Tree with probability of occurrence in blue circles

For ex: if “Low Fuel” and “Alarm Failure” causes “Seized Engine”, then it is represented using “AND gate”. If either “Starter Motor Failure” or “Battery Drained” causes the “Engine Not Starting”, then it is represented as “OR gate”. As you can see, any number of causes for a single event can be represented using logical gates. A single undesired event (product failure mode), can have multiple levels of causes with unique branches. The blue buttons on the bottom of each event describe its probability of occurrence. 

How will it help in RCA? Let’s say during your design phase, you have developed FTA for your product covering the most prominent failure modes. That is an excellent starting point to do RCA. Go down the tree and trim out the branches that do not support the observed data. The branch that is left standing is the probable root cause of the observed failure (undesired event). 

FTA with RCA 2
Sample Fault Tree with branches eliminated during RCA

“How can we incorporate FTA efficiently? Like everyone else, we are already overloaded with tasks.” Start building FTA when you start the RCA process. The benefits are tremendous when this is done in place of other traditional methods like fishbone. FTA will capture the linkage between causes, Fishbone will not. FTA will capture how your product’s environment & operator usage might combine to cause a failure. Doing RCA with FTA will also enable you to start assigning a numerical probability of each tree branch, enabling you to prioritize design improvement early. This creates efficiency in where you put your limited resources. In this way, FTA is a net gain in time with better results. List out five places in your program you could implement FTA right now. It is one of those amazing changes that yields better results and creates increased program efficiency.

Share this post